Q19 - Recent studies have demonstrated

[dan]

19. (A)
Question type: Identify a Flaw
The argument is flawed in that it concludes a correlation without sufficient justification. Certainly there could be positive correlation between caffeine and heart disease, but there could also be a negative correlation (caffeine prevents heart disease) or no correlation at all (they have nothing to do with each other). Answer choice (A) addresses this flaw, and is therefore correct.

(B) can lead to inferences that weaken the correlation between caffeine and heart disease, but is not significant enough to prove that such a correlation doesn’t exist at all.
(C) is not something that needs to be accounted for in reaching the conclusion.
(D) does not weaken the correlation.
(E) is the most attractive incorrect answer, because it represents a different relationship between caffeine and heart disease _ they may both be primarily caused by another factor. However, even if they are both caused by another factor, caffeine and heart disease could still impact one another. Therefore, (E) does not reveal a significant weakness within the conclusion.

[gyfirefire]

Could anyone help me with (E). i figured if there IS a common cause of both the development of heard disease and smoking and drinking caffeinated drink, then the original argument would be weakened that there is a positive correlation between drinking caffeinated beverages and development of heart disease.

Really appreciate it.

[dan]

This is a tough question, and a really good example of how the LSAT always keeps things fresh. Generally, we keep our eye out for correlation/causation issues. We know the pattern. The argument sets up a correlation, and then concludes a causation.

X is correlated with Y.
Therefore, X causes Y.

The flaw we know to look out for is:

Fails to take into consideration that there is something else, some third element, that causes both X and Y.

#19 sets us up for this -- it smells like a correlation/causation issue right from the start, and answer (E) seems to give us that third cause. However, the conclusion is NOT about causation. The conclusion is actually about correlation!

Answer (E) tells us that there is something that causes both heart disease and behaviors such as drinking caffeine and smoking. If there is a common cause for heart disease and drinking caffeinated beverages, then this would actually strengthen the conclusion that the two are correlated.

Make sense?

[geverett]

I fell for answer choice D. Would you agree that it only strengthens the authors premise that there is not a causal relationship between caffeine and heart disease?

[timmydoeslsat]

I do not like the fact that it says caffeine consumption vs caffeinated beverages.

It could be thought to strengthen the idea as long as we assume that it is not the case that all smokers do not have a hereditary predisposition to heart disease.

[ManhattanPrepLSAT1]

Hey Timmy!

Great explanations above! I think you're really got the reasoning down that underlies this argument.

I would like for an instructor to talk about the implications of a phrase change in answer choice D.

replace caffeine consumption with drinking of caffeinated beverages.

I think the phrase change in answer choice (D) is okay and I would eliminate it immediately - but it definitely offers a reason for elimination, if we ended up deciding between a couple of answer choices. The better reason to eliminate answer choice (D) is the one you mentioned above. It simply presents a flaw in concluding that drinking caffeinated beverages causes heart disease. And I love this answer choice, because so many of us are trained to jump on the correlation vs causation issue in a way that answer choice (D) would have been correct for. So in a way, it's the test writer responding to the instruction they know students have been receiving.

Here's my rule of thumb to deal with this issue section wide, since sometimes the correct answer requires being very loose with the language and sometimes it requires being very strict.

In the beginning of the section, I'd give them tremendous flexibility on language, unless I came to multiple answer choices, and then I'd become more strict to help me eliminate. Later in the section, I'd reverse that and begin being strict, but loosen it if I end up with no answer choices.

[robinzhang7]

Can someone go into a more in-depth explanation of correct answer choice A? Is it saying that there is NO positive correlation because smokers who do not drink CB are more likely to develop HD than smokers who do drink CB, that CB is really much less correlated with HD than originally thought.

From a numerical standpoint:

Smokers w/ CB: 0 cases HD
Smokers NO CB: 200 cases of HD

Then there is ABSOLUTELY no correlation between CB and HD?

Could someone take a look at this? Thanks!

[ohthatpatrick]

Yeah, you've nailed it.

Smoking often goes with caffeine and often goes with heart disease.

But that doesn't prove that caffeine and heart disease EVER go together, much less have a positive correlation.

(A) is simply providing a negative correlation, which works directly against the conclusion.

Per 100 smokers, let's say that 20 drink caffeine and 10 get heart disease.
Meanwhile,
Per 100 nonsmokers, let's say that 12 drink caffeine and 7 get heart disease.

That is enough to make the first two sentences true.

In order to judge the conclusion's truth value, you would need to fill in these blanks:
Per 100 caffeine drinkers, ________ get heart disease
vs.
Per 100 non-caffeine drinkers, _______ get heart disease.

If the conclusion is correct, that 1st blank will have a higher number than the 2nd one.

But (A) is saying that, at least within the population of smokers, the 1st blank has a LOWER number than the 2nd one.

[robinzhang7]

In order to judge the conclusion's truth value, you would need to fill in these blanks:
Per 100 caffeine drinkers, ________ get heart disease
vs.
Per 100 non-caffeine drinkers, _______ get heart disease.

If the conclusion is correct, that 1st blank will have a higher number than the 2nd one.

But (A) is saying that, at least within the population of smokers, the 1st blank has a LOWER number than the 2nd one.

I'm still slightly confused, partly maybe on how a positive correlation is defined in this problem. I thought a positive correlation meant the increase of variable X is accompanied by an increase in variable Y. Given your example, where is there an increase/decrease in any variable?

If I were to plug in numbers for the population of smokers:
Per 100 caffeine drinkers, 20 get heart disease
vs.
Per 100 non-caffeine drinkers, 30 get heart disease.

I'm still not sure what these numbers ^ explain about a correlation. In fact, can't these numbers show that there is a correlation between non-CB drinkers and HD and a separate correlation between CB drinkers and HD? I was looking to negate the conclusion's positive correlation with a negative one, something along the lines of.... an increase in CB is likely to lower HD. However, this problem is comparing smokers who drink CB vs. smokers who DON'T drink CB. These are two different groups. To disprove a correlation don't we need to focus on one group: say CB drinkers who are also smokers and their rates of heart diseases as a response to their different levels of CB drinking?

I was thinking along the lines of this:

Among smokers....
100 CB ----- 50 HD
0 CB ------- 70 HD

Which is exactly what answer A says? So the decrease from those who drink CB to those who don't (100-->0) is actually followed by an increase in HD (50-->70) which is what results in the negative correlation. What do you think, Patrick? This problem definitely caused some serious confusion for me.

[ohthatpatrick]

Get smoking out of your head for a sec, because the conclusion is making a statement PURELY about a positive correlation between CB and HD.

The conclusion is saying that, to at least some extent, the more CB ppl drink, the more HD they develop.

(A) is simply saying, among smokers (but it could really be among ANY population and it would still hold the same relevance to the conclusion), "the more CB drinking you do, the LESS HD you develop".

I think you're getting confused because you're trying to judge a correlation with sliding variables on a continuum.

But we're talking about a binary statistic here: you do or don't drink CB's and you do or don't develop HD.

So you're not going to be getting any LESS caffeine, LESS heart disease.

Moving from the caffeine-drinking population (where I was saying there was a 20% rate of HD) to the non-caffeein-driking population (where I was saying there was a 30% rate of HD), you're seeing LESS caffeine drinking going hand in hand with HIGHER risk of heart disease.

That IS the negative correlation you were hoping for that would basically go against the conclusion. (although, as an aside, we don't actually need to show there's a NEGATIVE correlation to go against the conclusion ... there is a middle ground, which is simply that there is no correlation between CB and HD ... that would also suffice to weaken the author).

[redskateboard]

The argument takes takes the invalid argument form:

A is correlated with B.
B is correlated with C.
Therefore A is correlated C.

Since the argument is invalid we cannot conclude A is correlated with C. The correct answer choice to this question catches that flaw.

[GingerB823]

Hi!

I'm still very confused about this one. I don't think I've ever felt more stuck about an answer that felt (and still feels) obvious and is apparently wrong.

A says smokers who drink CB are less likely to develop heart disease than are smokers who do not drink CB. This doesn't eliminate the positive correlation at all between CB and heart disease; it simply divides the group of smokers by whether they drink CB and tells us smokers who do not drink CB do not have as strong a positive correlation (if one at all) compared to those who drink CB. This still leaves lots of room for a positive correlation among smokers who drink CB, albeit one that is not as strong as smokers who do not drink CB.

D, however, seems to eliminate the positive correlation altogether between CB and heart disease. By telling us that only those caffeine drinkers with a hereditary predisposition to heart disease actually show a positive correlation with heart disease, B is saying those smokers who drink CBs and are not genetically predisposed show either no correlation or negative correlation to heart disease.

Thank you so much!

[jeanne'sjean]

The better reason to eliminate answer choice (D) is the one you mentioned above. It simply presents a flaw in concluding that drinking caffeinated beverages causes heart disease. And I love this answer choice, because so many of us are trained to jump on the correlation vs causation issue in a way that answer choice (D) would have been correct for. So in a way, it's the test writer responding to the instruction they know students have been receiving.

I'm still a bit confused by this one.

Does Timmy mean that if (D) includes some smokers into the group of "people who have a hereditary predisposition to HD", then the gap between prem (correlations with smokers) and conc (correlation between drinking CB and development of HD) is bridged, otherwise this answer choice will ONLY target at the CONC and even strengthen, not weaken it?

If an answer choice, if any, is directly AGAINST the conc but not the link between the premise and conclusion, will that answer choice can be a qualified weakener? I guess it can be. Since the weakener (or strengthener) can either zero in on the premise, the conclusion or the link between them.

When you say answer choice D is the test writer response to our training on correlation/causation issue, do you specifically mean your previous post as below?

X is correlated with Y.
Therefore, X causes Y.

The flaw we know to look out for is:

Fails to take into consideration that there is something else, some third element, that causes both X and Y.

#19 sets us up for this -- it smells like a correlation/causation issue right from the start, and answer (E) seems to give us that third cause. However, the conclusion is NOT about causation. The conclusion is actually about correlation!

Thanks a TON!

[MeenaV936]

Can someone answer Jeanne's post above? I'm wondering the same thing. IF this was a correlation vs. causation question, which one would be correct -- D, or E, or would both D and E be correct answers?

That being said, for correlation questions like this one, we should always doubt that the correlation exists in the first place, right? Just want a general principle to base my thinking/reasoning on for questions like these in the future. Thanks!

[ohthatpatrick]

If this were a typical correlation -> causality argument, it would sound something like this:

People who drink caffeinated beverages are more likely than those who don't to develop heart disease. Clearly, then, if you want to limit your risk of heart disease, you should limit your consumption of caffeinated beverages

This conclusion assumes that "drinking caffeinated beverages increases your risk of heart disease".

We'd react by thinking:
- what could be a different reason why caffeinated beverages and heart disease are correlated?

and

- what could increase/decrease the plausibility of the theory that drinking caffeinated beverages increases your risk of heart disease?

(D) could weaken the argument by saying, "the correlation exists only among people who have a hereditary predisposition to heart disease ... so it's probably not that caffeinated beverages increase your risk of HD; it's just that caffeinated beverages are connected to some third factor (genetic predisposition) that increases your risk of HD"

(E) could weaken the argument if it were saying "there is a common cause of both the development of heart disease and the behavior of drinking caffeinated beverages". That answer would ALSO find a different way to make sense of the correlation between caffeine and heart disease without blaming the heart disease on the caffeine.

========

As for the actual argument .... you were asking this:
That being said, for correlation questions like this one, we should always doubt that the correlation exists in the first place, right? Just want a general principle to base my thinking/reasoning on for questions like these in the future. Thanks!


I'm not sure what you mean by "correlation questions like this one".
If you mean "arguments in which the CONCLUSION is saying 'there must be a correlation'", then you should realize that this is one of only two questions ever that has done that. So it's not really a "thing" you're preparing for.

But, yes, we'd be doubting the correlation exists, because it's the author's conclusion.

For all arguments (in the Assumption family), the conclusion has not been proven. The argument is unconvincing. There's still wiggle room / still room for doubt.

I actively argue the "Anti-Conclusion" on most arguments:
"Given this evidence, how can I argue the conclusion is wrong?"

So I would have been thinking,
"Given that caffeine and smoking are correlated and that smoking and heart disease are correlated, how can I argue that caffeine and heart disease are NOT correlated?"

Hope this helps.